The WINDOWS NT operating system (or “WINDOWS NT”) from Microsoft Corporation of Redmond, Wash. provides a set of windowed utilities that allows easy setup and administration of a security system. The WINDOWS NT operating system itself is secure and makes its security system available to all applications through a standard Win32 security API.
An important aspect of the WINDOWS NT security system is that it is user-centric. Each line of code that attempts to access a secure object (file, printer, pipe, service, etc.) must be associated with a particular user. A user must identify himself to WINDOWS NT using a user ID and a password, via a log-on function. Each security check is made against the user's identification.
As a result, it is not possible, for example, to write code that prevents an application (e.g., Microsoft EXCEL) that is running under WINDOWS NT from accessing an object. For instance, an object can be secured against access from user Joe running EXCEL, but if user Carla is allowed to access the object, she can do so using EXCEL or any other application. All Carla has to do is identify herself to WINDOWS NT using her password.
Thus the entire validity of the WINDOWS NT security system is based on accurate identification of the user. WINDOWS NT user authentication is based on user IDs and passwords. Once a password is compromised, a general collapse of the security system can occur. There is therefore a need for a capability that adds a second factor to password-based authentication mechanisms such as that of WINDOWS NT. Such a capability should also ensure robustness while improving end-user convenience.
Not only do passwords present a security risk, they are also costly to administer. To provide an acceptable level of security, it is not uncommon to require changing corporate users' passwords every 30 to 60 days. This is not only an annoyance to the user, it is a major resource drain on system administrators. Surveys have shown that over 50% of the calls received by internal corporate hotlines are password related. Adding to this the lost productivity of professional office workers trying to figure out what their correct current password is, or requesting to be reinstated on the network, leads to an estimated annual cost of maintaining passwords of as high as $300 per user.
Saflink Corporation, with funding from the U.S. Department of Defense, has developed a Human Authentication application program interface (API), or HA-API, which allows applications to work with multiple biometric technologies presently available today and to integrate with new technologies in the future without requiring changes to the applications. The HA-API specification provides a set of standard program names and functions that enable various biometric technologies to be implemented easily into application programs for network user identification and authentication. It is foreseen that HA-API will be used both by application/product developers who wish to integrate biometric technology into their applications as well as by biometric vendors who wish to adapt their technologies for use within open system application environments.
FIG. 1 is a block diagram illustrating the architecture of HA-API. HA-API provides two interfaces. The first interface is an application API 101 consisting of functions 103 to determine which biometric technology (finger image, voice, facial image, etc.) is available to the application 10 and a set of functions 105 to authenticate a user's identity via any of the available technologies. The HA-API authentication functions 105 hide the unique characteristic of each biometric from the application 10. The second interface is a Biometric Service Provider (BSP) Interface 111 which provides a common interface for biometric technology providers to “plug-in” their unique modules 150. BSP modules 150 contain the capture, extraction (converting biometric features into a digital representation called a Biometric Identifier Record), and matching capabilities of a biometric vendor.
The full text of the Human Authentication API has been published by the Biometric Consortium (available at biometrics.org).